Browse Prior Art Database

TFR Expert System Shell

IP.com Disclosure Number: IPCOM000102425D
Original Publication Date: 1990-Nov-01
Included in the Prior Art Database: 2005-Mar-17
Document File: 4 page(s) / 149K

Publishing Venue

IBM

Related People

Bulko, WC: AUTHOR

Abstract

Because the task of configuring communications software can be very complex for some components, a study was initiated to determine whether or not artificial intelligence techniques could be applied towards simplifying the user interface. A basic expert system inference engine was needed for the development of prototypes, but most of the available ES shells were simply too large to use for a small component of a product where space was at a premium. TFR (Tiny Forward Reasoner) was created as a small, fast alternative to the bulky tools currently available. Although the original goal was to provide a dedicated ES shell providing only the functionality necessary for the intended product, TFR emerged as a versatile tool for developing expert systems applications.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 39% of the total text.

TFR Expert System Shell

       Because the task of configuring communications software
can be very complex for some components, a study was initiated to
determine whether or not artificial intelligence techniques could be
applied towards simplifying the user interface.  A basic expert
system inference engine was needed for the development of prototypes,
but most of the available ES shells were simply too large to use for
a small component of a product where space was at a premium.  TFR
(Tiny Forward Reasoner) was created as a small, fast alternative to
the bulky tools currently available. Although the original goal was
to provide a dedicated ES shell providing only the functionality
necessary for the intended product, TFR emerged as a versatile tool
for developing expert systems applications.

      TFR is a simple expert system shell for constructing rule-based
systems using forward chaining.  Its purpose is to allow the user to
develop the Presentation Manager* applications in an OS/2*
environment.  Written in the C language using the IBM C/2 compiler
and the OS/2 Programmer's Toolkit, its advantages include:
- a small size .  The most recent version of the executable for TFR
is less than 45K, remarkably small for an expert system tool.
- a simple, expandable design .  TFR provides six antecedent and
seven consequent clauses which satisfy most of the basic requirements
for a rule-based system.  The mechanism used to define clauses has
been designed to facilitate the addition of new expressions and
actions, if needed, by the programmer.
- a PM interface mechanism .  A key requirement of the TFR design was
to facilitate the prototyping of PM applications.

      The following pseudocode describes the top-level operation of
the TFR inference engine.
      MAIN:
         /*  initializations  */
         build the parameter list from the parameter file
         build the rule lists from the rule file
         if (the user specified a save file as a command-line
argument) restore the TFR state
         else
            load in any initial values for parameters
         /*  main select-execute cycle loop  */
         while not done {
            for each untriggered rule
               if all of its expressions evaluate to true add it to
the agenda
            if (the agenda is not empty) {
               select the rule whose rating evaluates the highest
execute the actions of the rule
            }
            else
               surrender
         }
         save work

      In order to define the TFR knowledge base, the user must first
create a PARMS.DAT file containing a list of parameter definitions,
one per line.  Parameters are user-defined internal variables,
represented within TFR using a multiple-linked list structure. They
can be sing...